1. Field of the Invention
Generally, the present invention relates to the formation of microstructures, such as advanced integrated circuits, and, more particularly, to the formation and non-destructive examination of conductive structures, such as metal regions, and their characteristics during stress conditions.
2. Description of the Related Art
In the fabrication of modern microstructures, such as integrated circuits, there is a continuous drive to steadily reduce the feature sizes of microstructure elements, thereby enhancing the functionality of these structures. For instance, in modern integrated circuits, minimum feature sizes, such as the channel length of field effect transistors, have reached the deep sub-micron range, thereby increasing performance of these circuits in terms of speed and/or power consumption. As the size of individual circuit elements is reduced with every new circuit generation, thereby improving, for example, the switching speed of the transistor elements, the available floor space for interconnect lines electrically connecting the individual circuit elements is also decreased. Consequently, the dimensions of these inter-connect lines have to be reduced to compensate for a reduced amount of available floor space and for an increased number of circuit elements provided per unit die area. The reduced cross-sectional area of the interconnect lines, possibly in combination with an increase of the static power consumption of extremely scaled transistor elements, may require a plurality of stacked metallization layers in order to meet the requirements in view of a tolerable current density in the metal lines.
Advanced integrated circuits, including transistor elements having a critical dimension of 0.13 μm and even less, may, however, require significantly increased current densities in the individual interconnect lines, despite the provision of a relatively large number of metallization layers, owing to the significant number of circuit elements per unit area. Operating the interconnect lines at elevated current densities, however, may entail a plurality of problems related to stress-induced line degradation, which may finally lead to a premature failure of the integrated circuit. One prominent phenomenon in this respect is the current-induced material transportation in metal lines, also referred to as “electromigration,” which may lead to the formation of voids within and hillocks next to the metal line, thereby resulting in reduced performance and reliability or complete failure of the device. For instance, aluminum lines embedded into silicon dioxide and/or silicon nitride are frequently used as metal for metallization layers, wherein, as explained above, advanced integrated circuits having critical dimensions of 0.13 μm or less, may require significantly reduced cross-sectional areas of the metal lines and, thus, increased current densities, which may render aluminum less attractive for the formation of metallization layers.
Consequently, aluminum is increasingly being replaced by copper and alloys thereof that exhibit a significantly lower specific resistivity and exhibit significant electromigration effects at considerably higher current densities compared to aluminum. The introduction of copper into the fabrication of microstructures and integrated circuits comes along with a plurality of severe problems residing in copper's characteristic to readily diffuse in silicon dioxide and a plurality of low-k dielectric materials and to have a moderately low adhesion to the low-k dielectrics. To provide the necessary adhesion and to avoid the undesired diffusion of copper atoms into sensitive device regions, it is, therefore, usually necessary to provide a barrier layer between the copper and the dielectric material in which the copper lines are embedded. Although silicon nitride is a dielectric material that effectively prevents the diffusion of copper atoms, selecting silicon nitride as an interlayer dielectric material is less than desirable, since silicon nitride exhibits a moderately high permittivity, thereby increasing the parasitic capacitances of neighboring copper lines. Hence, a thin conductive barrier layer that also imparts the required mechanical stability to the copper is formed to separate the bulk copper from the surrounding dielectric material and only a thin silicon nitride or silicon carbide or silicon carbonitride layer in the form of a capping layer is frequently used in copper-based metallization layers. Currently, tantalum, titanium, tungsten and their compounds with nitrogen and silicon and the like are preferred candidates for a conductive barrier layer, wherein the barrier layer may comprise two or more sub-layers of different composition so as to meet the requirements in terms of diffusion suppressing and adhesion properties.
Another characteristic of copper significantly distinguishing it from aluminum is the fact that copper may not be readily deposited in larger amounts by chemical and physical vapor deposition techniques, in addition to the fact that copper may not be efficiently patterned by anisotropic dry etch processes, thereby requiring a process strategy that is commonly referred to as the damascene or inlaid technique. In the damascene process, a dielectric layer is first formed which is then patterned to include trenches and vias which are subsequently filled with copper, wherein, as previously noted, prior to filling in the copper, a conductive barrier layer may typically be formed on sidewalls of the trenches and vias. The deposition of the bulk copper material into the trenches and vias is usually accomplished by wet chemical deposition processes, such as electroplating and electroless plating, thereby requiring the reliable filling of vias with an aspect ratio of 5 and more with a diameter of 0.3 μm or even less in combination with trenches having a width ranging from 0.1 μm to several μm. Although electrochemical deposition processes for copper are well established in the field of electronic circuit board fabrication, a substantially void-free filling of high aspect ratio vias is an extremely complex and challenging task, wherein the characteristics of the finally obtained copper metal line significantly depend on process parameters, materials and geometry of the structure of interest. Since the geometry of interconnect structures is determined by the design requirements and may not, therefore, be significantly altered for a given microstructure, it is of great importance to estimate and control the impact of materials, such as conductive and non-conductive barrier layers, of the copper microstructure and their mutual interaction on the characteristics of the interconnect structure so as to insure both high yield and the required product reliability. In particular, it is important to identify and monitor degradation and failure mechanisms in interconnect structures for various configurations so as to maintain device reliability for every new device generation or technology node.
One important aspect of copper-based lines and regions with respect to performance is the crystalline structure of the copper and copper alloys, since the effective resistance of the copper lines may depend significantly on the number, size and orientation of the crystal grains in the metal lines. Consequently, the materials used and also the processes used, such as deposition and post-deposition processes, may have a significant influence on the performance of these lines. Furthermore, any operational conditions may alter the crystalline structure, thereby also contributing to a performance degradation. Other degradation processes, such as stress-induced material transport, for example electromigration, may affect the crystallinity of the metal. Thus, great efforts are made in investigating the effect of metal grains on the overall performance of metallization structures, wherein, however, usually complex and destructive measurement procedures may be required. Consequently, a “direct” observation of process- and material-caused effects as well as operation-driven influences on the crystallinity of metal lines may be time-consuming, in particular, when the performance of a plurality of stacked metallization layers has to be evaluated in its entirety.
Since advanced microstructures, such as fast microprocessors, may require increasingly complex interconnects with dense metal structures at extremely reduced dimensions and many inspection techniques are already pushed to their limits, there exists a need for enhanced or alternative techniques, while avoiding or at least reducing one or more of the above-identified problems.